Some topological and mixed minimax theorems (Q5933721)

Minimax theorems are theorems about equality of maximin and minimax values under some assumptions. The well-known assumption is convexity-concavity of the objective function and convexity and compactness of the sets (the fundamental theorem of game theory or von-Neumann's theorem). In the paper, minimax theorems are proved under weaker assumption, in particular, the sets of constraints are not required to be compacts. The authors classify the conditions that are weak analogies of concavity and convexity. Their combinations have been used in minimax theorems by König, Ricceri, Simons, Cheng, Lin, Yu, Domokos, and Kindler. Two left combinations are involved in minimax theorems in the present paper. The authors use compactly locally upward and finitely weakly downward functions that they introduce. This gives a generalization of Simons' (Kindler's) minimax theorem which uses upward and (weakly) downward functions. Also this gives two new minimax theorems involving interval spaces and the following conditions: \[ f(x,y)\geq\min\{f(x^1,y),f(x^2,y)\}\;\forall y\;\forall x\in[x^1,x^2] \;\;\forall x^1,x^2; \] \[ f(x,y)\leq\max\{f(x,y^1),f(x,y^2)\}\;\forall x\;\forall y\in[y^1,y^2] \;\;\forall y^1,y^2; \] where \([x^1,x^2]\) and \([y^1,y^2]\) are connected subsets of interval spaces \(X\) and \(Y\).